Although combination antiretroviral drug therapies rapidly push plasma HIV-1 RNA concentrations below limits of detection, latently infected resting memory CD4+ T-cells can persist for years despite fully suppressive therapy. The intrinsic half-life of these cells has been estimated to be six months or longer. Recent studies suggest that populations of latently infected cell populations may persist for 60 years or more despite highly suppressive combination antiretroviral therapy. These long-lived infected cells, therefore, are thought to be the major impediment to curing patients of HIV-1 with antiretroviral drugs. However, they may not be the only reservoir for infectious virus: virus may also persist for long periods of time on follicular dendritic cells and in tissue macrophages. Current procedures for estimating infected cell half-lives are limited to tissues, such as PBMC, which can be sampled quantitatively over time. In this study I propose to use data on the decrease in genetic divergence (i.e., distance from the founder virus in patient) that occurs following antiretroviral therapy to draw inferences concerning the turnover of latently infected cell reservoirs. Declines in genetic divergence will be obtained using maximum likelihood trees. To quantify these declines I will develop mathematical models for viral dynamics that account for viral mutation and the decay of different infected cell compartments following antiretroviral therapy. Parameters for the diversification of virus in short-lived infected cells will be obtained from pre-treatment divergence data, while the input of virus into the latent reservoir will be modeled using viral load data. These models will be tested from published data and data obtained from experimental collaborators on HIV-1 sequences from infected children and adults who have gone on antiretroviral therapy. The estimates obtained from these models will be compared to estimates obtained from the decay in the density of latently infected cells following antiretroviral therapy. These models may give new insights into why therapies do not eradicate virus from the body and may give clinicians new ideas for strategies for targeting viruses in long-lived cellular compartments.